AWS Financial Services Symposium 2025 | Scott Mullins, AWS

Clips
News
More from AWS Financial Services Symposium 2025

Scott Mullins

Managing Director, Financial Services

AWS

Inside the AWS Financial Services’ AI playbook: 8 expert conversations from labs to ledgers

AI for financial services is no longer a distant vision — it’s reshaping how the industry thinks about risk, regulation, customer experience and innovation itself. From streamlining fraud detection to modernizing data architectures, financial institutions are leveraging AI not just to keep up, but to lead.As enterprise-grade tools mature and cloud platforms such as Amazon Web Services Inc. scale up their offerings, these organizations are deploying AI in tangible, transformative ways — and watching new value vectors emerge in real time.That momentum was on full display during the AWS Financial Services Symposium, where theCUBE captured a series of candid, expert-driven conversations

play_circle_outline Storytelling event with industry symposium format

play_circle_outline Maximizing Financial Productivity with Agentic AI Integration and Measurable Value in Services

play_circle_outline Organizations relying on past practices vs. adopting cloud technology

play_circle_outline Maximizing Value: Data Organization and Unified Studio for SageMaker Simplification in AI

play_circle_outline Revolutionizing Labor Spend: Startups Redefine Vertical Markets Through Innovative Software as a Service Models

Info
Transcript

Scott Mullins, AWS

Scott Mullins

Managing Director, Financial Services AWS

In this AWS Financial Services Symposium segment, Scott Mullins, managing director of financial services at AWS, joins theCUBE Research’s Dave Vellante and Scott Hebner for a discussion on how the cloud is reshaping the financial industry. The conversation dives into real-world progress as institutions modernize legacy systems, embrace generative AI and pursue scalable, secure digital transformation.

Mullins offers unique insights into how financial institutions are using generative AI via Amazon Bedrock to revitalize aging infrastructure, streamline c... Read more

explore Keep Exploring

What is the purpose of the financial services symposium and other industry symposiums organized by AWS? add

What were some examples of companies using agentic AI to improve productivity discussed in the keynote presentation? add

What are some challenges organizations face when trying to build a platform for experimentation and scaling in a safe way, and how can they overcome these challenges? add

What is the most important aspect of utilizing generative AI according to leaders in the field? add

What is the evolution of how we consume value from software today and how is business logic and value delivered to end users? add

bolt Powered by CUBE AI

Scott Mullins, AWS

search

Dave Vellante

>> Hi, buddy. Welcome back to New York City. You're watching theCUBE's continuous coverage of AWS Financial Services Symposium. I'm Dave Vellante with Scott Hebner and Scott Mullins is here as the managing director for Financial Services Worldwide at AWS. Hey, thanks for taking some time. Nice job today in the keynote. Looking good.

Scott Mullins

>> Well, keep going, I like it. But most importantly, thanks for being here, folks. We really appreciate the opportunity for you to be here, for you to tell the viewers at theCUBE what's happening at the New York Symposium. It's our 10th one, as I said from the stage, and we're just super excited about the opportunity to have everybody here today. It really was a storytelling event.

Dave Vellante

>> I just published my keynote analysis, go to thecuberesearch.com, and I loved how you were talking about, you started with 35 financial services pros at the Roosevelt Hotel, which I think is closed today. I mean, it was standing room only. The crowd, I mean, it was many, many hundreds. I don't know what the number is, five, 600, maybe even more?

Scott Mullins

>> Yeah, I think it's more than that.

Dave Vellante

>> Yeah, impressive. Maybe it as much as 700. I mean, it was really, really good. I liked a lot of the key messages. I especially liked, well, a couple things, you invoked Clark, which was really actually fantastic. And then I really liked the theme of blaze, break and push. They said, "Blaze new trails, break barriers and push beyond." And you had three guests on that talked about that, S&P Global, of course you guys always have the customers, S&P Global, Coinbase and Nasdaq, so I thought that mapped well to those themes, so congratulations on whoever that brainchild was.

Scott Mullins

>> Well look, thank you for that feedback and the compliments. What we try to do with our financial services symposiums and really all of our industry symposiums, it's a format that we use across industries at AWS now, is we try to take an opportunity to tell stories that aren't necessarily deeply technical in the keynote. And so you may have noticed, that in the keynote, you didn't hear me get really technical or see me throw up architectural designs on something because we're at a point, I think, where the vast majority of organizations understand the technical benefits of AWS and what it means to their business and the rate of innovation that we have around all the different services that we offer. Where we spend the majority of our time with financial institutions, is talking about change, from a people and a process perspective, an operating model perspective, that goes along with that and so hopefully the stories that you heard today, while they involve technology and the application of technology in business, they were about what it means for people to actually blaze new trails or break down existing barriers and processes and then push beyond what we have all thought were the typical traditional limits of what is possible and so it sounds like that resonated with you, so mission accomplished for us, I guess.

Dave Vellante

>> Well, it does only because, well, it was very clear to me you've moved beyond the whole message of cloud migration. I mean, even though your CEO, Andy Jassy and Matt talk about only whatever, 10%, 15% of the workload, that's still-

Scott Mullins

>> 15, 85% is still on-prem....

Dave Vellante

>> on-prem. Yeah, we can debate those numbers, but the majority of the workloads are still on-prem, so you're not really banging the drum so much on cloud migration. People understand that it's really about this business transformation, agentic adoption and the future capabilities with some customer proof points. That's what resonated to me, and I presume it was by design.

Scott Mullins

>> By design. I think we're at a point, in at least the financial services industry, where it's not about convincing people of the value of the technology, it's about helping people understand how they can move through that change, both the technical change and the underlying component change out of our services for legacy technology, but also the fundamental change that comes from your people and your processes.

Scott Hebner

>> Yeah, there was, just back to your point about getting value out of all this, I've doing a lot of work on this. I know MIT came out recently with a survey that only about 12% of businesses can actually measure the value of generative AI to their organization, and they want to start to get more measurable metrics as they go forward, so clearly there's been a benefit, there's been productivity, there's been value, but it's just hard to measure because it's more individual. As we move into this world of agentic AI and agents, how do you see that changing? Because you're getting into workflows and teams working together and you're elevating the objectives of what you're trying to achieve, right?

Scott Mullins

>> Well, hoping you saw some of that from some of our guests in the keynote today, right? And so I thought you saw two amazing stories from number one, S&P Global and then number two, from Coinbase. And so if you go back to Frank's presentation from S&P, and he talks specifically about their use of agentic AI and the way they've built agents and the improvements that they're seeing. He mentioned their productivity from a development perspective. He says, "It's right about 33% right now." And he said, "We're going to get it to 66%." And he repeated himself, he said, "Based on what we're seeing today, we are going to get it to 66." Because of number one, the capabilities they're able to bring to bear but number two, the fact that they're breaking down legacy processes. And he mentioned earlier, in his presentation, the challenge of process and how it's really not about the tool, it's about breaking down those processes. And then you heard Alex from Coinbase talking specifically about the fact that, from the chatbot, the agent that they had built that talks to Coinbase users, they were actually able to save five million people hours by not having to go out and hire people to meet peak demand for when the market was responding in a very positive way to cryptocurrency. And so I love the fact that we're at a point now, to your point, Scott, where we can actually show measurable value in the adoption of these technologies and the application to business.

Scott Hebner

>> Yeah, the ROI ramp is now going to start becoming more vertical because your getting agents, they're helping, like you said, they're digital workers, they're more than just an assistant doing something, they're helping you achieve goals, make decisions, and they're process level, so inherently you're going to be able to measure that much better, I think.

Dave Vellante

>> And then I think I also took a correct note that S&P Global shut down 17 data centers.

Scott Mullins

>> That's right.

Dave Vellante

>> I think that was the stat they used. And you're right, Coinbase slashing developer onboarding to, I think it was with a matter of weeks, and then you also cited Northwestern Mutual. They built this multi-agent evaluator to enforce PII guardrails. I thought that was interesting. Sort of an agent of agents, right?

Scott Mullins

>> Yeah.

Dave Vellante

>> Pretty cool.

Scott Mullins

>> Yeah. I mean look, it's one thing to build an agent to do something, it's another thing to make sure that you're checking that the agent is doing exactly what you wanted it to do in the first place, and so that was the point of the evaluator for Northwestern Mutual.

Dave Vellante

>> As always, AWS puts up a number of customer proof points. You had ADP, you had Bridgewater, you had Crypto.com, Rocket, Moody's, Robinhood, a number, I think there were nine or 10 on the slide. It's still early days for agentic adoption. Where are we in terms of, I mean, these are leading examples, there's kind of this fat middle that is still trying to, I think, figure it out, what are you seeing and what are you advising in financial services?

Scott Mullins

>> Well, I think we said it a little bit in the keynote presentation today. I'm happy to reinforce that. I think we've moved beyond the proof points of productivity. Early use cases were, "Let me take this technology, let me use it on myself. Let me see if I can actually make myself better and the processes that I have internally better." And so early adoption in 2023 and then in '24, as I mentioned, it was all about you, "How can I take this and actually do something that will generate value for ourselves?" And then some organizations kind of stopped on those rocks, right? They began to apply the technology, it was just a bolt on to existing processes and they didn't necessarily see value. Other organizations actually took the opportunity to actually begin to change processes, and you saw the outputs, like we saw from S&P Global and that you saw from Coinbase as well. Now, we're at a point where organizations are actually beginning to apply this in business. And so not just their own productivity, but actually enabling their existing line of business folks to leverage agents for business value. That may not necessarily be, "Hey, go and actually take an action for the company on behalf of a financial consumer. "But it is definitely, at this point, "Go and inform somebody, who's going to make a decision, in a way that can actually help them get to value quicker."

Scott Hebner

>> I really like where you guys are heading in what you're doing around this notion of decision intelligence, because I think the AI assistants with GenAI are more about analyzing information, creating content, identifying patterns. You start moving into the world of actually helping people make real decisions and be able to explain those decisions and the work you're doing with RAG, at different advanced levels, you're doing with knowledge graphs. I see you're doing some stuff with causal AI, I think those are the ingredients for this next wave to actually help AI make decisions.

Scott Mullins

>> Well, that's enabling organizations like Rocket and Rocket is presenting in a breakout today to improve the mortgage lending process, right? It's decisioning. When you make a decision about either mortgage lending or credit decisioning or underwriting, you're making a business decision. And so that's why you're seeing our focus there in helping organizations actually make much more intelligent decisions.

Dave Vellante

>> Bedrock is fundamental to AWS's AI strategy. Actually, it was quite clever, you had a picture of yourself, Nova generated this picture, you gave an example of the first time I had to file a claim yesterday and it was so simple, you were on your smartphone, out in front of your house, when the tree fell on your home. And you had mentioned, the same thing, I'm like, "It really resonated with me." Because the policy is on my wife's name and I had to go in and do that and it was so simple to do, so pretty amazing. I had dinner at AWS re:Invent with Rohit, who was the sort of person behind the Nova team, why is it important, you got Bedrock, all about choice, any model under the sun, open, closed, et cetera. Why is it important for AWS and Amazon specifically to have its own model? What are financial services customers telling you?

Scott Mullins

>> You said it, it's choice, right? So if you look at Amazon, fundamentally, Amazon was founded on a couple of basic principles. Number one, customers want choice, they want selection. Number two, they want that choice in selection at a really cost performant point, right? And then number three, they want it fast, so if you think about Amazon and the Amazon retail platform, when you order something, let's say you're ordering a period of tennis shoes or you're ordering something for your home, you want selection, you're going to be able to choose what's best for you, you want it at a really price performant point and you want it fast. And so when you think about our focus with Nova and when you heard Andy come out on stage at re:Invent last year and announced Nova, what did he hit? Probably the majority of his time on stage price, performance, price, performance. You can get performance that's equal to or better than other foundation models that are popular in the market today at a better price point. That's the name of the game, is making sure that we can offer our customers selection that meet their business needs. There's not one model that's ever going to fit every single use case, and so when you look at Nova and you look at the different iterations of the Nova models, we're actually giving customers the ability to use tools that make sense for their use case.

Scott Hebner

>> And that plays the idea of digital labor, right? Over time, you're going to be acquiring agents, you're going to be building your own, they may come from different suppliers, you're able to wire those things together and use the models that you want. So, I think the notion of being open and providing choice makes a lot of sense and the market's going to move there, right? And you're a couple steps ahead of that.

Scott Mullins

>> And the only way you're going to actually apply this in business is if it makes business sense. And what I mean by that is that it's got to be price, performance, cannot be spending money on things that aren't going to actually be accretive to your business. It can't just be an additional cost.

Dave Vellante

>> You mentioned Andy on stage, I observed, what I wrote at the time is it's almost like the Graviton playbook applied to LLMs. In other words, well, what does he say? "I've never heard a customer say they want to pay more." And so if you can get a more efficient model... I think, that's Bezos. We had Benioff on recently and he wrote an article in the Wall Street Journal saying, "I'm the last generation of managers that will be managing human only workforce. In the future, you're going to be managing humans and agents." First of all, do you buy that premise and what does that mean for how we will manage organizations in the future generally and specifically in financial services?

Scott Mullins

>> Well, I've not seen the article, so I'm going to have to go look at that because it sounds very interesting. I think the answer is we don't know right now, right? I think we talked a lot about today in the keynote about process change. You think about the way any organization runs today, whether it's your organization or our organization at AWS and Amazon or a financial institution, and you think about management and specifically we have processes that we use. We have procedures that we use. And so it's actually interesting, the point that Mark raised is like, "Hey, now we're going to have digital workers and we're going to have human being workers." It'd be interested to do the performance reviews on the digital workers. I think that'll be a little bit easier than some of the-

Scott Mullins

>> Indeed....

Scott Mullins

>> reviews on human ones, but it's posing a very interesting question on how work is done going forward. I think, at this point in time, it's all about agents enabling humans and what we were talking about earlier, which is driving more intelligent decision making. It's a lot easier to grade an agent from a performance perspective and actually make those adjustments and changes in real time than it's for a human being.

Dave Vellante

>> I asked Arvind Krishna a similar question last week, and he had a similar answer to what you just gave, Scott, which is, "Look, our job, it's not command and control, so it's not my job to sort of manage all these agents. It's my job to enable the humans to give them the tools to be able to take advantage." And that speaks to me to a more decentralized form of management. How does that square with financial services? My sense is it tends to be a lot of command and control. Is that maybe a misconception on my part? Is that changing, potentially?

Scott Mullins

>> Well, I think we can probably apply a command and control viewpoint of financial services just based on the obligations that financial institutions tend to have based on their regulatory obligations, specifically. And if you operate in even just one regime or location, you've probably got a number of different regulatory obligations based on the types of businesses that you're in. Maybe you're in the banking business, you're also in the securities business, you might also be in the insurance business. Well, guess what? That's probably three, four, six different regulators just here in the U.S. that you're dealing with, either at a federal level or a state level. If you happen to be a global organization and you're in one other jurisdiction, then multiplied by six, if you happen to be in 20 other jurisdictions, then you're at JP Morgan Chase or you're a Citigroup, right? And so, there can be a good degree of complexity when it comes to operating across environments like that. Adena Friedman at NASDAQ has a great term that I like. She talks about complexity versus complicatedness, and sometimes I think we can confuse the two. And so when we talk about this command and control, I think it's one of those things, right? Complexity doesn't have to be something that leads to complicatedness and leads to this, "I have to have a tops down command and control thing." And one of the wonderful things about just cloud adoption in general and not necessarily generative AI, is that we've been able to actually evolve the way that we think about risk management in the industry. Talking about the way that we build things, the way that we operate things, the way that we can help improve resiliency through architecting and building in different ways. And it's actually, I think, opening up the opportunities for innovation because we can de-risk the industry and the opportunity for change.

Scott Hebner

>> This reminds me of a conversation I had with Abdi Goodarzi, who's the head of GenAI at Deloitte, and he was saying that he's hearing from customers where the chief AI officer, obviously technical and business in nature, but has also becoming sort of the chief HR officer for digital agents. And the thought is someone has to govern these agents that get deployed because they're supposed to be little digital workers that pair up with people and help them make decisions. I thought that was really interesting because at the core of that is regulatory compliance policy, which is I know something you guys are really focused on-

Scott Mullins

>> It is something-...

Scott Hebner

>> especially in this industry.

Scott Mullins

>> It is something we're really focused on, Scott. I'm going to play a little coy here. It's something we're focused on because what we really want to make sure that we're doing is that we are enabling agents and organizations to have certified sources of truth. And so if you are going to empower an agent to make a decision about something, or if you're going to empower a chain of agents to make decisions based on different things that are happening, it's all going to be dependent upon what information they're being fed. And so that's an area that my team is looking at specifically is how do you ensure that an institution can rely on sources of truth that agents are pinging to make decisions against?

Scott Hebner

>> I was going to say, the last point is, it's a cultural thing too because the more people trust the agents, the higher the ROI, right? The worst thing that can happen is you build all these agents to help people and the people don't trust it, so I think the more, like you said, you build trust in the agents, it's through governance, compliance and-

Scott Mullins

>> Well, I'll give in. I'll give you a real life example from my team, right?. To your point, trust is based on performance in this agentic world, right? And so we, every year, process thousands of requests from financial institutions around due diligence, due diligence questionnaires or DDQs. We've done this for years, it's all part of the basic kind of governance process of, "Hey, I'm going to adopt new technology. Here's the control sets that I use. Can you please prove to me that you can follow those control sets and you can evidence them?" That process used to take us about a quarter, we'd get in a due diligence questionnaire from a financial institution, it's usually an Excel spreadsheet, it's usually got a number of tabs on it, there are a number of controls that maps different frameworks, whether that's NIST or COBIT or name your favorite popular framework for risk management. And then we would have to look at it, align our controls to it, show the evidence, and then hand it back. That was all manual mapping back in the day. We used machine learning over the course of several years to take that down from three months to one month. With the application of our generative AI tools, first using Bedrock and then using some of the popular foundation models, like Claude from Anthropic, and then now using our additional tools like Amazon Q for Business, we've taken that down to 10 minutes. So in 10 minutes, we can actually generate a response to a complicated due diligence question that used to take a quarter. Now, we're doing that with about an 80% accuracy rate, meaning, when it comes out, we have 80% confidence that it's correct when we give to a customer. That's not good enough for us.

Scott Hebner

>> No, right.

Scott Mullins

>> So, what do we do? We take an extra four hours now for a human being to go back and look at it. And so to your point, what we're now doing is continuing to improve that confidence score in the accuracy of what our agents that we've built for ourselves are going to be putting out, so they can again continue to lessen the time that our people spend doing that, when they can go and do other accretive work.

Dave Vellante

>> And you've dramatically reduced the elapsed time for the answer, and that job's going away, I'll tell you right now, that person's going to be happy. In the spectrum of optionality that you laid out from do-it-yourself to fully-managed agent support, are you discerning any patterns within financial services in terms of the preference? FS tends to have more skill sets, but then not all financial services companies are huge. What are you seeing in terms of the patterns that are emerging in terms of adoption along that spectrum?

Scott Mullins

>> Yeah, so to your point, Dave, it really depends on the organization, right? Some organizations have deep capabilities and some organizations may not have the resources at hand that a larger institution would have, so it really runs the gamut. Now, the wonderful thing about what cloud represents is the democratization of enterprise grade technology, right? It's no longer, "How much capital do you have on the balance sheet? What can you go buy and build for yourself?" It's, "Everybody has access to the same toolkit. Everybody has the access to the advantage of AWS by forming that relationship and then building through our tools." When it comes to this point, Dave, I think what you're seeing first and foremost is that organizations tend to rely on what they've done in the past. And so if you're an organization that in the past has been a build shop, guess what? You're building for yourself. In fact, we've had some instances where we've talked to customers and they said, "Hey, I'm really struggling. I'm trying to build this platform for my organization where I can actually bring in different foundation models and I can build guardrails around it, and I can enable this experimentation in a very safe and sound way, but it's not working. I'm having trouble scaling it." And the first thing I say to them is, "Why are you spending money on it? We've built Bedrock just for you." "Well, Scott, that's great. That's probably going to work for other organizations, but I am" insert the name of the company here "unique." And what I say back to them is that, "You're solving a problem that's already been solved. Let's go look at this together. Let's go apply it to your business." The same thing tends to happen with agents as well. "Hey, no, no, no, no, I've really got to do this myself." And the organization will put a bunch of money into trying to do this themselves, and then we'll be able to bring them proof points of, "Hey, have you tried this? Have you looked at this? This is scaling in other areas, this is working for other people." And then gradually you can see a transition to, "Okay, I don't need to do that. I can offload that undifferentiated heavy lifting and I can focus on value creation."

Dave Vellante

>> So many industries with heavy lifting. I mean, that original premise of AWS applies. One of the other things I observed in your keynote was like you said, you didn't go deep tech, you could, but this audience is more appropriate to talk about value above the core services. That said, when you think about compute storage and networking, there's a lot of innovation there. You think about what's happening with Trainium2, the evolution of Graviton and Nitro and Inferentia, you look at what's happening, we talked to Mai-Lan about things in data and storage and ultra-low latency networks, and you put up the slide, and I wrote this in my post infrastructure, "Follows the data." And then I put up your slide, "And so does our global infrastructure." And my question is around things like sovereign AI, with geopolitical tensions and other just nationalistic interests, you're seeing organizations and governments say, "No, we want to keep the data here." So, we're seeing this map of multicolored map of where you have AZs and regions and so forth.

Scott Mullins

>> It's like a Jackson Pollock painting at this point, right?

Dave Vellante

>> That's right. And I think there's a lot of confusion out there actually, because a lot of the competitors will say, "We have the most data centers anywhere." But we know, because we listen to Werner Vogels, about what it means to have resilient data centers and the ability to recover, but that underlying infrastructure and those core services remain a foundational. And now it seems to me you're really adding massive value on top of that, and that is happening at an accelerated pace, in many respects, because that foundation is there and continues to grow like the universe. I wonder if you could comment?

Scott Mullins

>> Well, we talked about a little bit in the keynote, right? I had the one slide that kind of had the iceberg of generative AI floating in the ocean, but really what was most important is what was underneath, what powers it all. And whether you listen to Mai-Lan or you listen to Dave Brown or any of our other leaders, they talk about the fact that if you really want to take advantage of generative AI, you've got to have a good solid foundation in data. Because if you don't have organized data, it's going to be very, very difficult to actually get value out of GenAI, so what's one of the things that we did at re:Invent last year? We announced SageMaker Unified Studio, so you don't even have to organize your data anymore. You can actually rely on SageMaker Unified Studio to actually make that a whole lot easier for you because think about how many organizations have struggled with their data estate. You've got data in different databases, maybe in a data lake that you built, and then maybe there was a lake house. Maybe there's a platform that you're using from a popular third party to try and make this easy, and it's not so easy as you thought it was going to be. How do we actually take, again, undifferentiated heavy lifting off of people to make it a lot easier to work with these things? And the way that we do that is through the continued investment in our global infrastructure. And so that is one of the foundational elements of AWS, is that, "Where you are today and where you need us to be tomorrow, we're going to be." Again, 36 geographic regions today, we've got another five that are on the way.

Dave Vellante

>> I love that Unified Studio announcement because SageMaker frankly was a heavy lift for a lot of organizations, and that goes a long way into simplifying.

Scott Hebner

>> Yeah, I've been talking to a lot of startups in this space of agentic AI and three things jump out. One is go vertical in a nuanced markets. Two is rethink software. I think to your point that it's not like building a SaaS app, you got so much infrastructure there to take advantage of, you can get things out much quicker. And the third thing was target labor spend, which is about 60% of the average (P&L) where the IT spend is five or 6%. And the notion was over time, these become truly digital workers, right? What are your thoughts on those three principles?

Scott Mullins

>> I haven't read that particular article, but that would be something that you'd be nodding north and south on because I think those are great points for any organization to be thinking about from the standpoint of, "How do I deliver value?" I had this conversation with Frank Tarsillo right before the keynote kicked off. We were talking about, "Hey, what's top of mind for everybody right now, in your organization at AWS, Scott, and for S&P?" And it's funny, Frank and I said the exact same thing at the same time, which is, "The evolution of how we consume value from what is software today." Right? Software as a service versus agentic value and so I think that focusing there, whereas how is business logic and business value delivered to end users, is a super critical area for all of us for the next two to five years.

Dave Vellante

>> We totally agree. In fact, we flipped the script on software as a service to services as software, and we're really trying to understand just like it was HCM or ITSM or pick your SaaS product, we're trying to understand what new services are going to be enabled and delivered as software as a result of agentic because we think, John Furrier has predicted it's going to be 10 x the size of the SaaS industry.

Scott Mullins

>> It's going to impact operating models, it's going to impact business models, it's going to completely fundamentally change the exchange of value between businesses. It's going to be rapid.

Dave Vellante

>> So, pay attention to what's happening in financial services because it's a leading indicator of what is possible in technology. Scott, thanks so much for coming on theCUBE.

Scott Mullins

>> Fellas, thanks for having us and thanks for being at the symposium with us this year.

Dave Vellante

>> You bet. It's our pleasure. And thank you for watching. Keep it right there. You're watching the Cube's coverage of AWS Financial Services Symposium, Dave Vellante for Scott Hebner. We'll be right back right after this short break. You're watching theCUBE.